Validation of the Japanese National Clinical Database Risk calculator for lung cancer surgery focused on postoperative morbidity.

OBJECTIVES To validate the efficacy of the Japanese National Clinical Database risk calculator, which predicts major morbidity in lung cancer surgery based on preoperative clinical characteristics. METHODS In total, 660 patients who underwent complete surgical resection of primary lung cancer were enrolled. The predicted rate of major morbidity determined using the risk calculator was compared between the patients with and without major morbidity. We performed receiver operating characteristic curve analysis to determine their cut-off values to predict major morbidity and assessed the associated factors with major morbidity. Major morbidity was defined as the Clavien-Dindo classification grade IIIa or greater. RESULTS The predicted rate of major morbidity was significantly higher in patients with major morbidity than in those without (P < 0.001). The cut-off value of the predicted rate of major morbidity to predict major morbidity was 3.0% (area under curve 0.741; sensitivity and specificity, 85.3% and 54.3%, respectively). The predicted rate of major morbidity ≥ 3.0% was significantly associated with occurrence of major morbidities (odds ratio 6.9; 95% confidence interval 2.63-18.04; P < 0.001) and the predicted rate of major morbidity had the highest odds ratio over other risk factors. This condition, namely the predicted rate of major morbidity ≥ 3.0%, was met in 315 (47%) of the total cases. However, only 29 (9%) of these cases had major morbidity in practice. CONCLUSIONS The risk calculator was fairly useful for estimating high-risk patients; however, it was not possible to identify a specific cut-off value to predict major morbidity in this cohort.